<p>In response to the “Dual Carbon” strategy, optimizing delivery routes is crucial for enhancing the efficiency of cold chain logistics enterprises. However, existing studies often simplify multi-objective conflicts into single-objective problems, and general optimization algorithms have limitations when solving discrete routing problems. To address this, this study proposes a multi-objective delivery route optimization method for cold chain logistics based on the Improved Grey Wolf Optimizer (GWO). This method introduces a discrete mechanism to address the route encoding problem, combines the Particle Swarm Optimization (PSO) algorithm to accelerate convergence in the later stages, and employs a linear crossover operation to enhance population diversity. This study constructs a three-objective model aimed at minimizing total cost, minimizing the number of time-window violations, and minimizing carbon emissions. Based on this, the study proposes the Multi-objective Grey Wolf Optimizer (Mo-GWO), which combines discrete encoding, hybrid initialization, diversity archiving, and hybrid search strategies, to solve the problem. Experimental results show that, compared to the Multi-Objective Particle Swarm Optimization (MOPSO) algorithm, the proposed Mo-GWO algorithm achieves an average 15.2% improvement in the supervolume metric of the Pareto solution set. Meanwhile, the total delivery cost is reduced by 8.9%, and the average vehicle loading rate increases to 85.7%. In the ablation experiments, the delete-insert operator had the most significant impact on algorithm performance; removing this module resulted in a total cost increase of 339 yuan. Therefore, this method can effectively improve cold chain logistics delivery efficiency, reduce enterprise operating costs, and decrease pollutant emissions.</p>